Kananaskis/Willmore Camera and Landcover Data

Taken from "Methods" in article: We used photographic data collected in remote camera arrays deployed using identical sampling designs and sampling methods (Heim et al. 2017) explicitly designed to be networked (sensu Steenweg et al. 2017) to examine predator distribution across gradients of disturbance the Canadian Rocky Mountains. Species were surveyed in a systematic grid-based sampling design, with individual 12 × 12 km2 grid cells as our sampling unit, the size of a wolverine’s home range which is the most wide-ranging species we sought to sample. Within grid cells cameras were placed on landscape features to maximize probability of detection. We deployed baited camera sites in both the WW (n = 66) and KR (n = 157) in October–December, monitoring monthly until March. Our sampling period is therefore representative of species’ activity over the prolonged winter season in the Canadian Rockies. Sites were accessed by ground (ski, snowmobile, all-terrain vehicles) or helicopter. While sampling years and duration differed between the two study areas (WW = 2006–2008; KR = 2011–2014), we have no evidence to indicate this affects any observed differences in species’ behaviour. In both areas Reconyx infrared-triggered digital cameras (models RM30, PM30 and PC900) were deployed paired with a frozen beaver carcass nailed to the tree facing the camera ca1 6-m away. Cameras were placed ca 1.5 m off the ground oriented north to prevent false triggers and set to high sensitivity with no trigger delay. Data were collected and bait replenished monthly (see Fisher and Bradbury 2014 for further details). We quantified human landscape disturbance using digital map inventories (Alberta Biodiversity Monitoring Institute, Human Footprint Map 2012). Linear features were the predominant landscape development feature present in the WW and the most widespread anthropogenic feature in the KR, so we used these as proxies of overall landscape disturbance. In the WW, seismic lines from petroleum exploration were the only linear feature (Fisher and Burton 2018) whereas in the more disturbed KR, this included a variety of anthropogenic linear disturbance features such as roads, recreational trails and industrial cutlines (e.g. seismic and power lines). Based on the available human footprint data, we calculated the percent cover (% area) of linear features at a 5000-m buffer around each camera site (Fisher et al. 2011) by merging all anthropogenic linear features into a single ‘linear disturbance’ variable.